Simulating and forecasting future scenarios
Energy modeling uses computer software to simulate the growth and function of energy systems in the real world. Models have been used for many years to plan, research and forecast how to react to different future scenarios. Recent advances in data science capabilities mean models can capture more complex and dynamic scenarios with increasingly sophisticated, flexible, and affordable results.
Models are objective simplifications of the real world and the choice of which assumptions and parameters are used can produce extremely varied results. Even the best models are imperfect representations of reality, with significant uncertainties. A common saying among modelers is “all models are wrong, but some are useful”.
This page owes a debt of gratitude to How Energy Modeling Works.
Energy modeling has three main categories of data:
- Existing energy system data
- Projections of future costs, policies, fuel prices, demand, and other data
- Various constraints, including technological, economic, political, and social justice
For example, a modeling process to identify the least-cost pathway to achieve net-zero emissions could incorporate ratepayer impact constraints for low-income residential customers. Models then use these inputs to generate outputs using complex mathematical optimization techniques.
The three primary types of energy model are:
- Capacity expansion models, which show how a system can change over time in response to policies, price changes, and technology trends that affect energy investments
- Production cost models, which show how a system will operate from day to day
- Power flow models, which simulate transmission and distribution networks to analyze how changes to the energy system impact the system itself
RESET teams mostly focus on the use of capacity expansion models.
All models have limitations, weaknesses, and biases.
Bad modeling can lock in long-term investments in undesirable strategies while overlooking opportunities to pursue desirable ones.
Poorly designed models are often based on flawed assumptions or low-quality data (‘garbage in, garbage out’), and do not have a proper awareness and assessment of uncertainties and risks. Many models assume perfect markets and planning, and optimize for system-wide conditions, even though individual actors make decisions based on their own interests and institutional constraints and pricing may mean that these interests do not align with those of society. Models can be used to promote or protect certain interests or technologies.
Proper interpretation of model results is just as important as a well-designed model. Neglecting to acknowledge the nuances or uncertainties of modeling can cause stakeholders to inappropriately interpret model outcomes.
The solution is to be transparent and honest in designing models and reporting their results.
Best practice for modeling is to lay out a range of scenarios based on differing assumptions and to indicate the varying degrees of uncertainty associated with the outcomes, as well as the key drivers of this uncertainty.
In 2013, NREL produced a report, containing ten key lessons for policymakers on modeling.
RESET strongly encourages members to make their energy models open-source and members work together to develop their skills in design and application of models, design and interpretation of scenarios and communication of results.